[ieee 2010 international conference on cyberworlds (cw) - singapore, singapore...

7
EEG-based “Serious” Games Design for Medical Applications Qiang Wang, Olga Sourina, and Minh Khoa Nguyen Nanyang Technological University Singapore e-mail: WANG0586 | EOSourina | [email protected] Abstract—Recently, EEG-based technology has become more popular in “serious” games designs and developments since new wireless headsets that meet consumer demand for wearability, price, portability and ease-of-use are coming to the market. Originally, EEG-based technologies were used in neurofeedback games and brain-computer interfaces. Now, such technologies could be used in entertainment, e-learning and new medical applications. In this paper, we review on neurofeedback game designs and algorithms, and propose design, algorithm, and implementation of new EEG-based 2D and 3D concentration games. Possible future medical applications of the games are discussed. Keywords- EEG; game design; HCI; BCI; e-learning; medical application; neurofeedback; fractal dimension I. INTRODUCTION “Serious” games usually have educational aims or health- related aims besides entertainment. Generally, such games could be classified as games for e-learning and medical applications. Recently, EEG-based technology has become more popular in “serious” games design and development since new wireless headsets that meet consumer demand for wearability, price, portability and ease-of-use are available in the market. Electroencephalogram (EEG) is a non-invasive technique recording the electrical potential over the scalp which is produced by the activities of brain cortex and reflects the state of the brain [1]. Different from other mental state interpreters, e.g. fMIR, EEG technique gives us an easy and portable way to monitor brain activities with the help of suitable signal processing and classification algorithms. Three main EEG-related application fields have been researched for several years: a) BCI applications that help disabled people to communicate with machines [2-3]; b) BCI applications for video games as game controllers [4]; and c) Neurofeedback games [5]. With the development of wireless EEG devices, expanding the EEG applications out of the lab became possible. In order to play EEG-based games, the user needs EEG reading cap/electrodes and computer/playstation. The software design consists of two main parts: signal processing algorithms and visualization/game. In work [6], we proposed to use fractal dimension (FD) model in neurofeedback games. In Section II A, review on EEG-based games for medical application is given. In Section II B, signal processing algorithms for EEG-based games are reviewed. In Section II C, game engines for such games development are described. Our game model, EEG processing algorithm and proposed games design are given in Section III. An implementation of the proposed 2D and 3D concentration games are discussed in Section IV. In Section V, conclusion and future medical applications of EEG-based games are discussed. II. RELATED WORKS A. Medical EEG-based “Serious” Games Neurofeedback is the technique that presents the real- time feedback to the user based on the EEG signals taken from the scalp of the user in the form of video display and/or sound [5]. Many researches reveal that the EEG and ERP (Event Related Potential) distortion always accompany psychological disorders such as Attention Deficit Hyperactivity Disorder (ADHD) [7-8], Autistic spectrum Disorders (ASD) [9-11], Substance Use Disorders (SUD) including alcoholics and drug abuse [12-13], etc. Similar to other parts of our body, the brain function can be trained as well as the EEG can be rectified. Neurofeedback (NF) is an alternative choice as a treatment to these disorders besides the medical treatments. Many neurofeedback games were assessed by the healing effect of the ADHD, one of the most known psychological disorders with significant EEG distortion. The θ/β ratio abnormal behavior was reported in [14]. Besides the ratio, the distortion in slow cortical potential (SCP) was also notified by [15]. Both the frequency neurofeedback training and the SCP neurofeedback training can achieve a good healing effect for ADHD [15]. ASD is a psychological disorder associated with abnormalities of social interactions and communications as well as serious restrictive interests and highly repeated actions [9]. In work [15], EEG analysis during eye open and resting condition was done for an eight-year-old girl with ASD patterns [16]. The α band and θ band of EEG signal act abnormally, and the corresponding neurofeedback scheme was designed to rectify abnormalities. After 21 sessions of treatment, the patient enhanced the sustained attention and decreased the ASD activities. Another research group also pronounced an achievement in neurofeedback treatment with standard Quantitative EEG (QEEG) protocol and aimed at decreasing theta band power at the central and frontal brain area [11]. Similar to ASD, General Anxiety Disorder (GAD) with which the patient cannot stop worrying can cause unacceptable social behaviors. GAD can also be treated with EEG α band suppression and symmetry training [17]. SUD including drug abuse or alcohol abuse always leads to disasters in social behaviors. Neurofeedback could be an 2010 International Conference on Cyberworlds 978-0-7695-4215-7/10 $26.00 © 2010 IEEE DOI 10.1109/CW.2010.56 270

Upload: minh-khoa

Post on 11-Dec-2016

213 views

Category:

Documents


1 download

TRANSCRIPT

EEG-based “Serious” Games Design for Medical Applications

Qiang Wang, Olga Sourina, and Minh Khoa Nguyen Nanyang Technological University

Singapore e-mail: WANG0586 | EOSourina | [email protected]

Abstract—Recently, EEG-based technology has become more popular in “serious” games designs and developments since new wireless headsets that meet consumer demand for wearability, price, portability and ease-of-use are coming to the market. Originally, EEG-based technologies were used in neurofeedback games and brain-computer interfaces. Now, such technologies could be used in entertainment, e-learning and new medical applications. In this paper, we review on neurofeedback game designs and algorithms, and propose design, algorithm, and implementation of new EEG-based 2D and 3D concentration games. Possible future medical applications of the games are discussed.

Keywords- EEG; game design; HCI; BCI; e-learning; medical application; neurofeedback; fractal dimension

I. INTRODUCTION

“Serious” games usually have educational aims or health-related aims besides entertainment. Generally, such games could be classified as games for e-learning and medical applications. Recently, EEG-based technology has become more popular in “serious” games design and development since new wireless headsets that meet consumer demand for wearability, price, portability and ease-of-use are available in the market.

Electroencephalogram (EEG) is a non-invasive technique recording the electrical potential over the scalp which is produced by the activities of brain cortex and reflects the state of the brain [1]. Different from other mental state interpreters, e.g. fMIR, EEG technique gives us an easy and portable way to monitor brain activities with the help of suitable signal processing and classification algorithms. Three main EEG-related application fields have been researched for several years: a) BCI applications that help disabled people to communicate with machines [2-3]; b) BCI applications for video games as game controllers [4]; and c) Neurofeedback games [5]. With the development of wireless EEG devices, expanding the EEG applications out of the lab became possible.

In order to play EEG-based games, the user needs EEG reading cap/electrodes and computer/playstation. The software design consists of two main parts: signal processing algorithms and visualization/game. In work [6], we proposed to use fractal dimension (FD) model in neurofeedback games.

In Section II A, review on EEG-based games for medical application is given. In Section II B, signal processing algorithms for EEG-based games are reviewed. In Section II C, game engines for such games development are described.

Our game model, EEG processing algorithm and proposed games design are given in Section III. An implementation of the proposed 2D and 3D concentration games are discussed in Section IV. In Section V, conclusion and future medical applications of EEG-based games are discussed.

II. RELATED WORKS

A. Medical EEG-based “Serious” Games Neurofeedback is the technique that presents the real-

time feedback to the user based on the EEG signals taken from the scalp of the user in the form of video display and/or sound [5]. Many researches reveal that the EEG and ERP (Event Related Potential) distortion always accompany psychological disorders such as Attention Deficit Hyperactivity Disorder (ADHD) [7-8], Autistic spectrum Disorders (ASD) [9-11], Substance Use Disorders (SUD) including alcoholics and drug abuse [12-13], etc. Similar to other parts of our body, the brain function can be trained as well as the EEG can be rectified. Neurofeedback (NF) is an alternative choice as a treatment to these disorders besides the medical treatments.

Many neurofeedback games were assessed by the healing effect of the ADHD, one of the most known psychological disorders with significant EEG distortion. The θ/β ratio abnormal behavior was reported in [14]. Besides the ratio, the distortion in slow cortical potential (SCP) was also notified by [15]. Both the frequency neurofeedback training and the SCP neurofeedback training can achieve a good healing effect for ADHD [15].

ASD is a psychological disorder associated with abnormalities of social interactions and communications as well as serious restrictive interests and highly repeated actions [9]. In work [15], EEG analysis during eye open and resting condition was done for an eight-year-old girl with ASD patterns [16]. The α band and θ band of EEG signal act abnormally, and the corresponding neurofeedback scheme was designed to rectify abnormalities. After 21 sessions of treatment, the patient enhanced the sustained attention and decreased the ASD activities. Another research group also pronounced an achievement in neurofeedback treatment with standard Quantitative EEG (QEEG) protocol and aimed at decreasing theta band power at the central and frontal brain area [11]. Similar to ASD, General Anxiety Disorder (GAD) with which the patient cannot stop worrying can cause unacceptable social behaviors. GAD can also be treated with EEG α band suppression and symmetry training [17].

SUD including drug abuse or alcohol abuse always leads to disasters in social behaviors. Neurofeedback could be an

2010 International Conference on Cyberworlds

978-0-7695-4215-7/10 $26.00 © 2010 IEEE

DOI 10.1109/CW.2010.56

270

affordable and considerable alternative treatment in SUD. Chronic alcoholics showed significant diminution in α band of EEG signals. Corresponding neurofeedback treatment can decrease the brain waves in these bands and showed an effect of such alcoholic treatments [12]. For drug abuse, decreased α band power was also found as well as the excess of fast beta band activities. In addition, a subject with drug addiction has lower amplitude in P300 ERP component than that of a controlled subject. The addiction can be relieved with long term neurofeedback treatment [13].

Besides medical applications, neurofeedback can also help a healthy person to enhance the brain functions. Researchers indicated that cognitive performances, e.g. cued recall performance, can be enhanced if a healthy person learns to increase special components of EEG signals with neurofeedback [18-19].

B. Algorithms The signal processing algorithms used in BCI system

especially in neurofeedback systems can be generally classified into two main methods, i.e. frequency analysis and event related potential (ERP) analysis. Frequency training is the most prevalent method in clinical applications together with the QEEG protocol.

EEG signal can be divided into several different frequency bands, i.e. δ band (<4Hz), θ band (4-7Hz), α band (8-12Hz), β band (12-30 Hz) and γ band (>30 Hz). Specially, the Sensorimotor rhythm activity (12 – 15 Hz) is also used in several neurofeedback systems. Each frequency band is related to different brain functions. Generally, δ band is prevalent in infant’s EEG or EEG when the subject is sleeping; θ band is prevalent in EEG when the subject feels drowsiness; α band is significant when the subject is relax; β band is associated with fast activities and γ band is related to problem solving and memory work [20]. The power over different bands were assessed and extracted from the patient EEG signals and then compared to the QEEG database (QEEG protocol) or statistical analysis was run to generate the pathology and corresponding recovery protocols. The frequency training method is the most prevalent method used in the neurofeedback training systems and other EEG applications because the frequency band power is easy to be obtained and analyzed with the existing signal processing tools.

ERP analysis is the process analyzing the EEG signal synchronized with an event. Slow Cortical Potential (SCP) and P300 are the important ERP approaches used in the neurofeedback treatment. SCP reflects the changes in cortical polarization, i.e. negative and positive trends, of EEG signals which last from 300ms to several seconds after event stimulus [21]. Abnormalities in SCP of ADHD patient were studied in [15], and the corresponding neurofeedback protocol could enhance the continuous performance. The P300 component of ERP occurs during 300ms – 600 ms after event stimulus which is obtained by oddball paradigm in which low-probability target items are inter-mixed with high-probability non-target items. Researches indicated that the amplitude of P300 component is related to the process of allocation of attention resources and its latency reflects the

stimulus evaluation and classification time. The pathology of P300 component in drug abuse patients was reported in [13], and neurofeedback based on P300 component training was proposed.

Although the signal processing algorithms embedded in neurofeedback games are well applied in clinical treatments, the linear features (power spectral density or amplitude) extracted from EEG cannot represent the brain activities perfectly due to the nonlinearity of EEG signal itself. Nonlinear method, e.g. entropy analysis and fractal dimension analysis, become popular in many EEG processing for medical applications [22-25] and could be applied to neurofeedback systems to model brain activities. In this paper, we proposed and described 2D and 3D concentration games based on the fractal dimension algorithms.

C. Game Engines EEG-based games consist of two parts: signal processing

and game implementation. Game implementation can be effectively done with the help of game engines. Game engines are tools that programmers use to design and implement games. They provide ready-made utilities or tools to develop a game. According to Jeff Ward [26], three types of game engines are frequently seen: roll-your-own game engines, mostly-ready game engines and point-and-click engines.

Roll-your-own game engines, including OpenGL and DirectX, require game makers to be well-versed in programming and take a lot of time to build. However, they give the game makers flexibility and more freedom in building their own components for the game. Mostly-ready game engine is most popular in the market. Renderer, physics engine, collision detection, graphic, sound system, etc. are usually available in these game engines. OGRE, Panda3D, Unreal, etc. belong to this kind of game engine. A point-and-click engine is the highest level game engine that requires least programming knowledge. However, they are quite limited with the ready-made functions provided. These engines include for example, Alice and Game Maker. As EEG-based games include signal processing, a game engine should support programming language C++, Python or any other scripting environments to allow EEG recognition and interpretation.

III. EEG-BASED GAME DESIGN

A. EEG-based Game Design Till now traditional neurofeedback games were

implemented for clinical applications with complicated EEG devices which are hard to set up. Recently, more and more portable and wireless EEG devices became available. More effective EEG processing algorithms which can be applied with fewer electrodes in real-time applications are on demand. In this paper, we propose design of neurofeedback concentration games based on fractal dimension model. Different from traditional neurofeedback games, we focus on the brain state monitoring of the EEG activities using fractal dimension model. The main idea of concentration game is

271

using neurofeedback for encouraging the concentration state by giving positive feedback if it is recognized from the EEG signals. The EEG analysis method is based on fractal dimension model which can capture the changes in brain state.

In order to develop real-time application, fewer channels and faster signal processing algorithms are expected. In our implementation, only one channel located in occipital lobe is selected because the occipital lobe is responsible for visual perception and visual attention [27-28]. Entropy based fractal dimension model [29] is used to distinguish brain states such as relaxed and concentrated. Then, an immersive game should be designed and behave according to the recognized brain state.

B. Fractal Dimension Model The most important novelty in this implementation of the

neurofeedback game is using the fractal dimension feature instead of the power of EEG signals. Fractal dimension is the measurement of complexity and irregularity of a signal. Higher fractal dimension values mean that the signal is more complex, while lower fractal dimension values mean that the signal is more regular. In Fig. 1 (a) and (b), examples of mono-fractal Weierstrass signals with low FD value 1.1 and high FD value 1.7 are shown.

For our real-time implementation, Higuchi [30] and Box-counting [31] algorithms were chosen for FD calculation.

In Higuchi method, the samples are first clustered into several sub-sequences according to the poly-phase structure shown in (1). The length of the sequences is calculated according to the (2). The denotes the number of the sub-sequence and denotes the length of the m-th sub-sequence. The total length is proportional to , where is the fractal dimension value and is the time-delay information. The fractal dimension calculation process of Higuchi method is represented in Fig. 2.

: , , 2 , , / (1) Lm k 1k |x m+ik -x m+ i-1 k |int N-mk

N-1

int N-mk k

1 (2)

Box-counting method can calculate the fractal dimension

value of the signal in time domain without any sub-sequence extraction steps. The main step of box-counting method is box construction. Unified and normalized boxes are constructed in 2D space (time-amplitude) which cover one segment of the signal. Finally, the number of boxes is counted. The boxes’ constructing and counting processes are shown in Fig. 3. The number of counted boxes is

proportional to , where denotes the length of the side of the boxes.

Figure 1. Mono-fractal Weierstrass signal. (a) Fractal dimension value is

1.1. (b) Fractal dimension value is 1.7.

Higuchi method and Box-counting method are compared in both the computation complexity and the accuracy. Brownian Motion and Weierstrass mono-fractal signals for which theoretical fractal dimension values are known were used for the comparison. Although Higuchi method is slower than Box-counting method, the accuracy of Higuchi method is better than Box-counting method as it is shown in Fig. 4 on Brownian Motion and Weierstrass signals. In our work, both algorithms are used for neurofeedback implementation.

C. Experiment on Brain States Classification An experiment on brain state classification was set up to

distinguish relaxed and concentrated states. Five subjects aged from 22 to 30 were invited to participate in the experiment. In the first session, in order to induce relaxed state, a comfortable environment was set up to help the subject relax. In the second session, in order to induce concentration state, the subjects were required to complete several math problems. Only one electrode was used and placed in O1 position according to 10-20 international system [32] in occipital lobe which is associated to the visual perception. EEG signals were recorded by the Emotiv device with sampling frequency of 128Hz and 16-bit A/D

272

resolution. EEG signal was processed by fractal dimension algorithms, and average fractal dimension values of both two sessions were evaluated for all five subjects.

Figure 2. Fractal dimension calculation in Higuchi method.

Figure 3. Boxes construction in box-counting method, the dark boxes are counted.

The results of the brain state recognition experiment are shown in Fig. 5 with boxplot. The states of relaxation and concentration have different fractal dimension values for all five subjects. In both Higuchi and Box-counting algorithms, the experiment results show that concentration level can be distinguished for 80% of the subjects when a default threshold was set to 1.9 in Higuchi method and 1.55 in Box-counting method. For 100% of the subjects, the concentration level can be recognized with a trained threshold. It is clear that fractal dimension model can be used for distinguishing the relaxation state and concentration

state with a simple threshold even though there are some overlap which may be due to the individual difference in the level of concentration. A short training session can be applied to determine the default threshold to minimize the individual effects.

Because this is not a clinical experiment, it is difficult to make a comparison between effectiveness of our methods with that of traditional neurofeedback games which are based on frequency analysis. The comparison will be arranged in the pain management experiments with the help of the Tan Tock Seng Hospital in the future.

Figure 4. The comparison of Higuchi and Box-counting algorithms for

FD evaluation over (a) Brownian Motion signals and (b) Weierstrass signals.

273

Figure 5. The comparison of (a) Higuchi method anmethod in FD evaluation of the EEG signals in differe

subjects.

IV. GAMES IMPLEMENTATI

In our research, one 2D neurofeedba“Brain Chi” and one 3D neurofeedbac“Dancing Robot” are developed for concontrol. “Brain Chi” is developed with engine, while “Dancing Robot” is develope[34] game engine. Both games are develoC++.

A. Data Acquisition and Processing EEG data is recorded by Emotiv [35]

device. Only O1 electrode (according to 10system) in Occipital lobe is active, and tlocation is transmitted to computer with algorithm filters out the unrelated componeband-pass filter first, then calculates the fvalue of the input EEG signals in real-timmethod or Box-counting method and labelsbrain states according to the adaptive threshothreshold used for distinguishing the concenrelaxation state is set to 1.9 in Higuchi mebox-counting method. The threshold cmanually or can be trained in a 20s trainingthe setting compatible to individuals. The and processing algorithm diagram is shown hardware setup of our neurofeedback game 7.

Read data

from Emotiv

FractalDimensionAlgorithm

AdaptivThreshoCalculati

Bandpass Filter(2-42 Hz)

Figure 6. The data acquisition and processing alg

PlayerEmotiv

EEG Device

Wireless Communication

Figure 7. Hardware setup of the gam

nd (b) Box-counting ent brain states for all

ION ck game named

ck game named ncentration level SDL [33] game

ed with Panda3D oped with Visual

EEG collection 0-20 international the EEG in this

Bluetooth. The ents with 2-42Hz fractal dimension me with Higuchi s it with different olds. The default ntration state and

ethod and 1.55 in an be changed

g session to adjust data acquisition in Fig. 6 and the is shown in Fig.

ve old ion

GameCommand

gorithm diagram.

Desktop Computer

n

me.

B. Game Strategy In our games, if the concentrati

are rewarded to the player to concentrate. If the brain state of thnegative feedback comes in to take The game strategy can also be chawhen the player wants training fostrategy diagram is shown in Fig. 8.

Figure 8. Game stratege for concen

“Dancing Robot” is a simple game, the player is required to cowhile the robot is dancing. A screenis shown in Fig. 9. Its dancing concentration level of the player. headset concentrates to make the robaccompanied by excited music. sluggishly if the player is distracted.employed for training’s purpose.

“Brain Chi” is a 2D single-playscreenshot of “Brain Chi” is showcontrols the game simply by using concentration. His/her task is to hagainst evil bats using a protectioprotection ball is controlled by the cplayer. To win this game, the playball by concentration to eliminate al

V. CONCLUSION AND FUTURE WIn this paper, we review

neurofeedback games, algorineurofeedback games and game efficiency of fractal dimension varesults of the experiment show thaBox-counting method can be eqfeature extraction methods. The brawith the difference in fractal dimenand 3D EEG-based games: “BraRobot” were designed and implemtraining. Fractal dimension methodneurofeedback games to enhanclassification algorithms.

There are also some aspects of enhanced in the future. For exampthreshold in the game, we can mavalue to concentration level dire

ion state is found, points encourage him/her to

he subject is distracted, a the player’s points away.

anged to relaxation game or relaxation. The game

ntration neurofeedback.

single-player 3D. In this ontrol the speed of robot nshot of “Dancing Robot”

speed depends on the The player with his/her

bot dance faster, which is The robot will dance

. An adaptive threshold is

yer EEG-based game. A n in Fig. 10. The player his/her “brain power” of

help a little boy to fight on ball. The size of the concentration level of the yer needs to increase the l the bats.

WORK ed EEG-based games ithm embedded in engines. Experiment on alue was arranged. The at both the Higuchi and

qually effective used as in states are recognizable nsion value. Original 2D ain Chi” and “Dancing mented for concentration d was embedded in the nce the efficiency of

the project which can be ple, instead of using one ap the fractal dimension ectly to achieve precise

274

control and training over the concentration level. On the other hand, clinical experiments should be arranged in the future to do the comparison between fractal dimension approach with the traditional neurofeedback games’ signal processing approaches.

Short clips of “Brain Chi” and “Dancing Robot” games can be found by visiting the following link:

http://www3.ntu.edu.sg/home/EOSourina/

Figure 9. “Dancing Robot” screenshot.

Figure 10. “Brain Chi” screenshot.

ACKNOWLEDGMENT This project is supported by grant NRF2008IDM-

IDM004-020 “Emotion-based personalized digital media experience in Co-Spaces” of National Research Fund of Singapore.

REFERENCES [1] P. L. Nunez and R. Srinivasan, Electric Fields of the Brain, 2

ed.: Oxford University Press, 2006. [2] B. Rebsamen, E. Burdet, C. Guan, H. Zhang, C. L. Teo, Q.

Zeng, et al., "A brain-controlled wheelchair based on P300 and path guidance," 2006, pp. 1101-1106.

[3] B. Rebsamen, C. L. Teo, Q. Zeng, M. H. Ang Jr, E. Burdet, C. Guan, et al., "Controlling a wheelchair indoors using thought," IEEE Intelligent Systems, vol. 22, pp. 18-24, 2007.

[4] A. Lécuyer, F. Lotte, R. B. Reilly, R. Leeb, M. Hirose and M. Slater, "Brain-computer interfaces, virtual reality, and videogames," Computer, vol. 41, pp. 66-72, 2008.

[5] D. C. Hammond, "What is neurofeedback?," Journal of Neurotherapy, vol. 10, pp. 25-36, 2006.

[6] Q.Wang, O.Sourina and M.K.Nguyen, "Fractal Dimension Based Algorithm for Neurofeedback Games," in Proc. CGI 2010. SP25, Singapore, 2010.

[7] J. F. Lubar, M. O. Swartwood, J. N. Swartwood and P. H. O'Donnell, "Evaluation of the effectiveness of EEG neurofeedback training for ADHD in a clinical setting as measured by changes in T.O.V.A. scores, behavioral ratings, and WISC-R performance," Biofeedback and Self-Regulation, vol. 20, pp. 83-99, 1995.

[8] T. Fuchs, N. Birbaumer, W. Lutzenberger, J. H. Gruzelier and J. Kaiser, "Neurofeedback treatment for attention-deficit/hyperactivity disorder in children: A comparison with methylphenidate," Applied Psychophysiology Biofeedback, vol. 28, pp. 1-12, 2003.

[9] R. Coben, M. Linden and T. E. Myers, "Neurofeedback for autistic spectrum disorder: A review of the literature," Applied Psychophysiology Biofeedback, vol. 35, pp. 83-105, 2010.

[10] L. Thompson, M. Thompson and A. Reid, "Neurofeedback outcomes in clients with Asperger's Syndrome," Applied Psychophysiology Biofeedback, vol. 35, pp. 63-81, 2010.

[11] M. E. J. Kouijzer, H. T. van Schie, J. M. H. de Moor, B. J. L. Gerrits and J. K. Buitelaar, "Neurofeedback treatment in autism. Preliminary findings in behavioral, cognitive, and neurophysiological functioning," Research in Autism Spectrum Disorders, vol. 4, pp. 386-399, 2010.

[12] E. Saxby and E. G. Peniston, "Alpha-theta brainwave neurofeedback training: An effective treatment for male and female alcoholics with depressive symptoms," Journal of Clinical Psychology, vol. 51, pp. 685-693, 1995.

[13] T. M. Sokhadze, R. L. Cannon and D. L. Trudeau, "EEG biofeedback as a treatment for substance use disorders: Review, rating of efficacy, and recommendations for further research," Applied Psychophysiology Biofeedback, vol. 33, pp. 1-28, 2008.

[14] R. J. Barry, A. R. Clarke, S. J. Johnstone, R. McCarthy and M. Selikowitz, "Electroencephalogram θ/β Ratio and Arousal in Attention-Deficit/Hyperactivity Disorder: Evidence of Independent Processes," Biological Psychiatry, vol. 66, pp. 398-401, 2009.

[15] H. Gevensleben, B. Holl, B. Albrecht, D. Schlamp, O. Kratz, P. Studer, et al., "Distinct EEG effects related to neurofeedback training in children with ADHD: A randomized controlled trial," International Journal of Psychophysiology, vol. 74, pp. 149-157, 2009.

[16] J. D. Cowan and L. Markham, "EEG biofeedback for the attention problems of Autism - A case study," Biofeedback and Self-Regulation, vol. 19, pp. 287-287, Sep 1994.

[17] C. Kerson, R. A. Sherman and G. P. Kozlowski, "Alpha suppression and symmetry training for generalized anxiety symptoms," Journal of Neurotherapy, vol. 13, pp. 146-155, 2009.

[18] D. Vernon, T. Egner, N. Cooper, T. Compton, C. Neilands, A. Sheri, et al., "The effect of training distinct neurofeedback protocols on aspects of cognitive performance," International Journal of Psychophysiology, vol. 47, pp. 75-85, 2003.

[19] S. Hanslmayr, P. Sauseng, M. Doppelmayr, M. Schabus and W. Klimesch, "Increasing individual upper alpha power by neurofeedback improves cognitive performance in human subjects," Applied Psychophysiology Biofeedback, vol. 30, pp. 1-10, 2005.

275

[20] J. N. Demos, Getting Started with Neurofeedback. New York: WW Norton & Company, 2005.

[21] N. Birbaumer, "Slow cortical potentials: Plasticity, operant control, and behavioral effects," Neuroscientist, vol. 5, pp. 74-78, 1999.

[22] V. Kulish, A. Sourin and O. Sourina, "Analysis and visualization of human electroencephalograms seen as fractal time series," Journal of Mechanics in Medicine and Biology, vol. 6, pp. 175-188, Jun 2006.

[23] V. Kulish, A. Sourin and O. Sourina, "Human electroencephalograms seen as fractal time series: Mathematical analysis and visualization," Computers in Biology and Medicine, vol. 36, pp. 291-302, Mar 2006.

[24] O. Sourina, A. Sourin and V. Kulish, "EEG Data Driven Animation and Its Application," in Computer Vision/Computer Graphics Collaboration Techniques, Proceedings, 2009, pp. 380-388.

[25] O. Sourina, V. V. Kulish and A. Sourin, "Novel Tools for Quantification of Brain Responses to Music Stimuli," in 13th International Conference on Biomedical Engineering,Proceedings, New York, 2009, pp. 411-414.

[26] J. Ward. (2008, What is a Game Engine? . Available: http://www.gamecareerguide.com/features/529/what_is_a_game_.php

[27] Z. H. Li, C. D. Coles, M. E. Lynch, X. Y. Ma, S. Peltier and X. P. Hu, "Occipital-temporal Reduction and Sustained Visual Attention Deficit in Prenatal Alcohol Exposed Adults," Brain Imaging and Behavior, vol. 2, pp. 39-48, Mar 2008.

[28] S. O. Murray and E. Wojciulik, "Attention increases neural selectivity in the human lateral occipital complex," Nature Neuroscience, vol. 7, pp. 70-74, 2004.

[29] W. Kinsner, "A unified approach to fractal dimensions," in Proc. ICCI 2005: Fourth IEEE International Conference on Cognitive Informatics, 2005, pp. 58-72.

[30] T. Higuchi, "Approach to an irregular time series on the basis of the fractal theory," Physica D: Nonlinear Phenomena, vol. 31, pp. 277-283, 1988.

[31] A. Block, W. Von Bloh and H. J. Schellnhuber, "Efficient box-counting determination of generalized fractal dimensions," Physical Review A, vol. 42, pp. 1869-1874, 1990.

[32] R. W. Homan, J. Herman and P. Purdy, "Cerebral location of international 10-20 system electrode placement," Electroencephalography and Clinical Neurophysiology, vol. 66, pp. 376-382, 1987.

[33] Simple Directmedia Layer. Available: http://www.libsdl.org [34] Panda3D. Available: http://www.panda3d.org. [35] Emotiv. Available: http://www.emotiv.com

276